Add debug logging to track issue_type saving

app.py

@@ -519,6 +519,13 @@ def save_label(decision, issue_type="", notes=""):
     traj = current_trajectories[current_idx]
     traj_id = traj.get("id", "")
     
+    # Debug logging
+    print(f"💾 save_label called:")
+    print(f"  decision: {decision}")
+    print(f"  issue_type: '{issue_type}' (type: {type(issue_type)}, len: {len(str(issue_type))})")
+    print(f"  notes: {notes}")
+    print(f"  traj_id: {traj_id[:20]}...")
+    
     row_data = {
         "dataset_repo": traj.get("dataset_repo", ""),
         "config_name": traj.get("config_name", ""),
@@ -530,19 +537,27 @@ def save_label(decision, issue_type="", notes=""):
         "timestamp": pd.Timestamp.now().isoformat()
     }
     
+    print(f"  📋 row_data issue_type: '{row_data['issue_type']}'")
+    
     # Upsert: update if exists
     existing_mask = evaluations_df['trajectory_id'] == traj_id
     is_update = existing_mask.any()
     if is_update:
         idx = evaluations_df[existing_mask].index[-1]
+        print(f"  🔄 Updating existing row at index {idx}")
         for col, val in row_data.items():
             evaluations_df.at[idx, col] = val
+        print(f"  ✅ After update, issue_type = '{evaluations_df.at[idx, 'issue_type']}'")
         status_msg = f"✅ Updated: {decision}"
     else:
+        print(f"  ➕ Adding new row")
         evaluations_df = pd.concat([evaluations_df, pd.DataFrame([row_data])], ignore_index=True)
+        new_idx = evaluations_df.index[-1]
+        print(f"  ✅ After add, issue_type = '{evaluations_df.at[new_idx, 'issue_type']}'")
         status_msg = f"✅ Added: {decision}"
     
     save_evaluations()
+    print(f"  💾 Saved to CSV")
     
     if dataset_sync_enabled:
         from hf_dataset_sync import append_to_dataset
@@ -556,9 +571,24 @@ def save_label(decision, issue_type="", notes=""):
 def save_with_decision(decision, review_reason, notes):
     # Ensure issue is always empty string unless decision is "review" with valid reason
     valid_choices = ['too_short', 'too_long', 'wrong_description', 'task_already_completed', 'mislabeled_success', 'mislabeled_failure', 'mislabeled_suboptimal', 'other']
+    
+    # Debug logging
+    print(f"🔍 save_with_decision called:")
+    print(f"  decision: {decision} (type: {type(decision)})")
+    print(f"  review_reason: {review_reason} (type: {type(review_reason)})")
+    print(f"  notes: {notes}")
+    
     issue = ""
     if decision == "review" and review_reason and str(review_reason) in valid_choices:
         issue = str(review_reason)
+        print(f"  ✅ Setting issue_type to: {issue}")
+    else:
+        print(f"  ❌ Issue NOT set. Checks:")
+        print(f"     decision == 'review': {decision == 'review'}")
+        print(f"     review_reason truthy: {bool(review_reason)}")
+        if review_reason:
+            print(f"     review_reason in valid_choices: {str(review_reason) in valid_choices}")
+    
     return save_label(decision, issue, notes)